1. Field of the Invention
This invention relates to improvements in a focus detecting apparatus for detecting the focus state from the relative positional relation between the two images of an object.
2. Related Background Art
As one type of the focus detecting apparatus of a camera, there is known the so-called "image deviation type" in which the exit pupil of the photo-taking lens is divided. The relative positional displacement of a pair of images formed by light beams passed through the pupil areas is observed to thereby determine the in-focus state.
The focus detecting method in the apparatus of this type will hereinafter be described with reference to FIG. 15 of the accompanying drawings. A field lens FLD is disposed coaxially with a photo-taking lens FLNS whose focus is to be detected. Disposed behind the field lens and at positions symmetric with respect to the optic axis are two secondary imaging lenses FCLA and FCLB. Further rearwardly of the secondary imaging lenses, there are disposed sensor rows SAA and SAB to sense visible or invisible images. Diaphragms DIA and DIB are provided near the secondary imaging lenses FCLA and FCLB. The field lens FLD substantially images the exit pupil of the photo-taking lens FLNS on the pupil surfaces of the two secondary imaging lenses FCLA and FCLB. As a result, light beams entering the secondary imaging lenses FCLA and FCLB are those emitted from regions of an equal area on the exit pupil surface of the photo-taking lens FLNS which correspond to the secondary imaging lenses FCLA and FCLB and do not overlap each other. When aerial images formed near the field lens FLD are re-imaged on the surface of the sensor rows SAA and SAB by the secondary imaging lenses FCLA and FCLB, the two images on the sensor rows SAA and SAB change their positions on the basis of the displacement of the positions of the aerial images in the direction of the optic axis. Accordingly, if the displacement (deviation) of the relative position of the two images is detected, the focus state of the photo-taking lens FLNS can be known.
The signal processing method for detecting the amount of image deviation from the image signals output from the sensor rows SAA and SAB is disclosed in Japanese Laid-Open Patent Application No. 142306/1983 (corresponding to U.S. Pat. No. 4,559,446), Japanese Laid-Open Patent Application No. 107313/1984 and Japanese Laid-Open Patent Application No. 101513/1985 (corresponding to U.S. Pat. No. 4,618,236). Specifically, when the number of picture elements constituting the sensor row SAA or SAB is N and the image signals from the first (i=0, . . . , N-1) sensor rows SAA, SAB are A(i) and B(i), the following equation is operated with respect to k.sub.1 .ltoreq.k.ltoreq.k.sub.2 : ##EQU1## M is the number of operation picture elements represented by (M=N-.vertline.k.vertline.-), k is called the amount of relative displacement, and k.sub.1 and k.sub.2 are often assumed as -N/2 and N/2. The operation max{a, b} represents the extraction of the greater one of a and b, and the operator min[a, b]represents the extraction of the smaller one of a and b. Accordingly, the terms X.sub.1 (k), X.sub.2 (k), Y.sub.1 (k) and Y.sub.2 (k) in equations (1) and (2l) above can be regarded as the amounts of correlation in a broad sense. Further, examining equations (1) and (2) in detail, X.sub.1 (k) and Y.sub.1 (k) in reality represent the amounts of correlation in (k-1) displacement as defined above, and X.sub.2 (k) and Y.sub.2 (k) represent the amounts of correlation in (k+1) displacement. Therefore, the evaluated amount X(k) which is the difference between X.sub.1 (k) and X.sub.2 (k) means the amount of variation in the amount of correlation of the image signals A(i) and B(i) in the amount of relative displacement k.
The amounts of correlation X.sub.1 (k) and X.sub.2 (k) are smallest when the correlation between the two images is highest, as is apparent from the above definition. Consequently, X(k) which is the amount of variation therein should be "0" and the inclination thereof should be negative when the correlation is highest. However, X(k) is scattered data and therefore, actually, on the assumption that EQU X(kp).gtoreq.0, X(kp+1)&lt;0 (3)
and that the peak of the amount of correlation exists in the section [kp, kp+1] of relative displacement wherein X(kp)-X(kp+1) is greatest, by effecting an interpolating operation ##EQU2## the amount of image deviation PR below the picture element unit can be detected.
Conversely to X.sub.1 (k) and X.sub.2 (k), the amounts of correlation Y.sub.1 (k) and Y.sub.2 (k) become greatest from the foregoing definition when the correlation between the two images is highest. Consequently, Y(k) which is the amount of variation therein should be "0" and the inclination thereof should be positive when the correlation is highest. As regards Y(k) also, like X(k), when EQU Y(kp).ltoreq.0, Y(kp+2)&gt;0 (6)
and Y(kp)-Y(kp+1) is greatest, by effecting an interpolating operation of ##EQU3## the amount of image deviation PR below the picture element unit can be detected.
Also, any of the focus evaluated amounts X(k) and Y(k) can be used to detect the amount of image deviation, but as seen from Japanese Laid-Open Patent Application No. 101513/1985, better accuracy in terms of S/N is obtained if the amount of image deviation PR is found by the use of the focus evaluated amount X(k) when .vertline.X(kp)-X(kp+1).vertline.&gt;.vertline.Y(kp+1)-Y(kp).vertline., and by the use of the focus evaluated amount Y(k) when .vertline.X(kp)-X(kp+1).vertline.&gt;.vertline.Y(kp+1)-Y(kp).vertline..
Now, in the correlative operation of equations (1) and (2), the number of operation picture elements M is calculated by the equation M=N-.vertline.k.vertline.-1, and the greater the amount of relative displacement k, the smaller the number of operation picture elements M. This is because as the amount of relative displacement k becomes greater, the output signals of the corresponding sensor rows SAA and SAB lack from the end. Thus, if the amount of relative displacement k becomes great when the number of picture elements N forming the sensor rows SAA and SAB is small, the number of operation picture elements M will become remarkably small and the S/N of the focus evaluated amounts X(k) and Y(k) will be deteriorated.